Ensign Nemo wrote on 2023-10-19, 22:37:
I've heard of putting bad HDDs in the freezer, but never to heat them. Interesting.
For laptops in the early 2000s sadly it's a necessity, the Toshiba xxxxGAS / xxxxGAX 2.5" drives all have fluid dynamic bearings and that fluid goes bad over time - I can get most of these GAX / GAS drives working temporarily enough to image just by heating the bearing up with a hair-dryer for a little bit, topping up the temperature if it stops responding.
One nice thing I can say for the regular ball bearing drives is they are less likely to fail from the bearing fluid going bad over time.
Thermalwrong wrote on 2023-10-18, 01:03:Putting together a full 486 system where I want 16MB of RAM, but it's got 2x 72-pin SIMM slots and I appear to have run out of 8 […]
Show full quote
Putting together a full 486 system where I want 16MB of RAM, but it's got 2x 72-pin SIMM slots and I appear to have run out of 8MB / 16MB SIMMs that I'm willing to spare for this system.
So experiments with making EDO memory function in a computer that has no support for its timings - today for the first time I tried with SoJ memory that's got both upper and lower CAS. It seemed to work but digital audio and music was horribly messed up when that stick up top was fitted.
The OR gate is my attempt to connect the OE (output enable) pin to UCAS and LCAS so that if both are low, OE is also pulled low but if either one goes high then OE also goes high / inactive. Messy but after fixing soldering errors it works with the airwires, not working right still though.
EDO-to-FPM-conversion-tests.jpg
Gave up on that one for now and picked up another 16MB EDO memory stick that got close to POSTing before being modified. This PCB is a nice layout, there's a clear trace connecting all of the OE pins to ground at the center point. I first cut that center point and after used a tiny drill to break the trace between the OE trace on each module, then soldered OE to the CAS pin which is next to it on each chip. Instead of trying to cut the traces with a knife after I cut an adjacent trace and my finger. The tiny drill bit proved to be pretty good for this task.
Verified that none of the modules are still bridged except for where the CAS lines join up, now this modified EDO 16MB module is working on this board and the sound is normal.
So far I've only had success modifying EDO into FPM on these 4-bit DRAM chips, rather than the 16-bit with UCAS/LCAS.
Finding as well that despite these being 50ns chips, the timings have to be pretty lax on this Abit AN4 to get even HIMEM to pass its memory test - I found the same on some other 4MB EDO sticks, can't really do better timings than good FPM modules.
Hmm, doing more testing with this Abit AN4 with the EDO > FPM converted SIMM - this converted 16MB stick is working at the "fastest" timings allowed in the BIOS with 40MHz bus speed (overclocking a DX2/66 to 80MHz) with the 1 wait state on DRAM writing. So maybe it is better than regular FPM in some ways - these chips are 50ns so they should be good.
While searching for some of the BIOS strings I found this impressively written and concise manual for a different 486 motherboard: https://theretroweb.com/motherboards/s/advanc … -1.11#downloads
It's got the recommended timings listed with pinouts, resulting frequencies from dividers around page 30. I'm tempted to print out some parts of it as a handbook for future reference 😀
Testing out some games at 80MHz, I was getting some really weird graphical corruption in games where the pixels looked okay but the screen tint kept changing or in doom it was going kind of 'all-blue'. This is the fun of VLB, I moved the video card up 1 slot closer to the CPU and now it's behaving again.